Train my own dataset，miss some targets

xingyizhou / CenterNet

Object detection, 3D detection, and pose estimation using center point detection:

MIT License

7.27k stars 1.92k forks source link

Train my own dataset，miss some targets #391

Open 1170300906 opened 5 years ago

1170300906 commented 5 years ago

训练后能够很好地框出一些目标，框的大小也非常正确，但是有一些目标没能框出来？我需要在参数上进行改正吗？

1170300906 commented 5 years ago

It seems that the threshold is 0.5? How can I change the threshold?

qh332612340 commented 5 years ago

you can change the threshold in opts.py

1170300906 commented 5 years ago

you can change the threshold in opts.py

which option?

1170300906 commented 5 years ago

How can I change the threshold for the center points' detection confidence so that it can produce more bounding boxes?

sailyung commented 4 years ago

k value; max_per_image value

ustczhouyu commented 3 years ago

Training and testing on my own data, many objects are missed, but when each image is cut into two images and then tested, there is no missed objects. Maybe it is because my data contains too many objects. How to modify the object number of output ?